Defence defence defence

[garret9]

Best way to get guys who are not on the ice for goals against?
Get guys who are not on the ice for shots against...

 

[Mathmew Purrrr Oh]

Couldn't agree more, Dax. I think that the new school thinking is that a guy who doesn't produce offensively or push the play offensively is not playing defense well.

Of course you don't want to be Stuart who is just blocking shots and trying to hit, running around your zone. However, I feel that there are very effective defensemen who really don't contribute much from the offensive blueline in.

I feel like I am a very good defensive defenseman. I am not often on the ice for goals against. I am good at retrieving the puck in my zone, forcing turnovers on my side of centre. I get my stick in lanes, separate opponents from the puck when they attack my zone. I have good gap, especially at the blueline. I can settle down the play in my own zone, reset, and I make a really good first pass and can skate the puck out of my zone.

I don't think I'd be liked as a defenseman around here though. I am too low event. I also block shots which I know is somehow a bad thing here because it means you don't have the puck

So you're Chris Tanev. Great. He's actually an excellent defensive defenceman and not just a mischaracterized ****** one.

 

[Mortimer Snerd]

On the Roslovic/Larrson debate.

The most favorable projection numbers I had, had Larson at a 33% success rate with an extremely small sample (3 guys)

With a 20 player sample (sort of my minimum) he was at 20%.

Roslovic on the otherhand was at over 70% success rate in a similar sized sample.

His lowest cohort % I got was 45% (due to having to widen the parameters due to very few players scoring at his rate in the USHL)

His closest comparables across NHLE's, Height, and Birthdate were Matt Duchene, Mike Richards and Claude Giroux.

His closest comp across height and league scoring was Paul Statsny.

There was a lot of evidence to say Roslovic was easily a superior pick.

Fair enough.
Do you bias value by position?
How good is your database to compare the USHL to the SHL and Swedish junior leagues Larsson played in? I have a hard time with Euro players at draft time because they typically play for 4-5 different teams in their draft years.
I biased Larsson for having played in a men's pro league in his draft year. Am I overvaluing that?
If you went back and re-evaluated with D+1 data would you come to the same conclusion?

 

[Mortimer Snerd]

We go around in circles. They were not equal on draft day on the list that counts. I doubt they are equal now on that list. I also disagree that we " know" what you claim we do. I don't know it. I trust the Jets evaluation of talent level more than yours. You are not going to convince me otherwise.

Everything you argue seems to be promoting positional drafting.

You are right. We are going around in circles. You seem to be willing to trust the Jets list without question.

I don't believe in positional drafting. Neither do I believe that anybody's evaluation is able to pick the absolute best prospect out of a group of 3-4 players taken 1 after the other in the draft after the first few. Even there we often see 3 or 4 players bunched so closely together that no one can really know which one is the best. Even among the experts you get many different opinions, all well thought out and defended by fact. They are looking at 17 YO kids and trying to identify who will be the best players 5-6 years later. So at each pick take the best player in the context of those available and of the team you are trying to build. Position is a factor 2 ways. Cs and Ds are more valuable than wingers and when you are loaded with Fs you should take the D out of that group of players who are so closely ranked that you really can't identify the best one. If one is clearly better than the others take him.

If I had the 1st OA this year I would not take a D. Mathews is clearly the best player bet. Laine is probably 2nd. After that it begins to get muddy.

Just as I can't convince you to draft positionally (not that I'm trying to), you can't convince me that any team's scouts can accurately put the top 30 players in order from 1 - 30. The error bars are huge. So when well within the error bars consider position along with several other factors.

Do you think anyone could look back at a draft 7 or 8 years ago and come up with a definitive list of the top 30 players from that draft, in order? That would be a list that every knowledgeable person would agree with.

 

[Mortimer Snerd]

Giving everything a relative value compared to a 1st overall pick being set at 100, the average drop of 3rd to 8th gives an additional value of 18.8.

This would work out to some combinations like:
- 25th
- 35th + 45th
- 94 7th round picks

I'm not sure I am following this correctly. Do you mean that in exchange for a 3rd we should get an 8th + a 25th? ........ Or alternatively an 8th + 94 7ths?

 

[Grind]

Fair enough.
Do you bias value by position?
How good is your database to compare the USHL to the SHL and Swedish junior leagues Larsson played in? I have a hard time with Euro players at draft time because they typically play for 4-5 different teams in their draft years.
I biased Larsson for having played in a men's pro league in his draft year. Am I overvaluing that?
If you went back and re-evaluated with D+1 data would you come to the same conclusion?

the 33% is almost built exclusively off the fact he's playing a man's league, and is where you get the most favorable projection. Considering a similar excercise with Roslovic (going for the most favorable projection) you get over double the liklihood of success i'd say yes, your definitly overvaluing that.

They both use NHLE's for the base projections which do decent job projecting to the NHL, but have issues particularly with young players in euro pro leagues as there is no TOI adjustment and these players are generally getting far fewer minutes then their junior level counterparts.

Furthermore, most players transitioning form SHL to NHL are older, so while it does have some issues, I think that's cleaned up pretty well with the favorable projection of 33%.

Also keep in mind this basing Defencemen progress off scoring points, which is less then perfect, but still has a decent success rate.

There are obvious imperfections, but there's a LOT of ground those imperfections would have to cover up to make it close, as Roslovic projected 2X as likely to succeed through pretty much every set of criteria I used. I don't think you can explain away a gap like that.

As for the D+1 season, i'll have to take a look. It's not done yet but 14 points 41 SHL games is not likely to have closed the gap considerably but will definitly be worth revisiting at the end of this year.

 

[Mortimer Snerd]

Jets are 7th last in overall goal%.
Jets are 8th best in overall goal% when leading, just one place ahead of the Chicago Blackhawks.
Jets are 3rd best in overall goal% when leading in the third period.

Jets are actually have been one of the better teams when leading.

Interesting. That contradicts a pretty widely held perception I think. I wonder if it varies significantly by goalie? Pavs certainly seems to give up leads a lot. That may be a misperception based on the big event. I'm pretty confident that the Jets do badly at coming from 1 goal behind. The number of EN goals we give up should seal that one.

 

[Mortimer Snerd]

I snipped some of your post but your numbers do provide a compelling case for choosing Ros over Larsson. But if Larsson had been at like 68% I think it would have been better to go with him even if Ros was BPA by a hair. Of course, if ifs and buts were candies and nuts, my boss would be one of these:

The bold has been my point. If I am mistaken about their relative values then my whole case falls apart. Wouldn't be the first time I have been wrong.

 

[Mortimer Snerd]

Couldn't agree more, Dax. I think that the new school thinking is that a guy who doesn't produce offensively or push the play offensively is not playing defense well.

Of course you don't want to be Stuart who is just blocking shots and trying to hit, running around your zone. However, I feel that there are very effective defensemen who really don't contribute much from the offensive blueline in.

I feel like I am a very good defensive defenseman. I am not often on the ice for goals against. I am good at retrieving the puck in my zone, forcing turnovers on my side of centre. I get my stick in lanes, separate opponents from the puck when they attack my zone. I have good gap, especially at the blueline. I can settle down the play in my own zone, reset, and I make a really good first pass and can skate the puck out of my zone.

I don't think I'd be liked as a defenseman around here though. I am too low event. I also block shots which I know is somehow a bad thing here because it means you don't have the puck

You describe a pretty good defenseman.

I think the bolded misstates the issue though. It seems to me that there is a misunderstanding between those who prefer that type of D man and those who prefer the modern offensive type. The offensive D man is presumed to be weak defensively allowing more goals against. He compensates for this weakness by scoring more goals as well. While that may happen sometimes I don't think that is the point or the objective. I think the idea is that the offensive D man has more puck possession which equals fewer SA against, therefore fewer goals against not more. He also scores more goals increasing his goal differential further. To take it a step further he also gets more O zone time which also reduces goals against and increases goals for. If he has fewer blocks it may be because he gives up fewer SAs. If he is weak at some of the more traditional defensive play he is making up for it by all those other things.

Lastly I don't think it is necessarily correct to say that either style is superior. It is the end result that matters and that is determined by the net goal differential at the end of the game. If one tends to win by 2-1 and the other tends to win 4-3 they are equal.

 

[Mortimer Snerd]

the 33% is almost built exclusively off the fact he's playing a man's league, and is where you get the most favorable projection. Considering a similar excercise with Roslovic (going for the most favorable projection) you get over double the liklihood of success i'd say yes, your definitly overvaluing that.

They both use NHLE's for the base projections which do decent job projecting to the NHL, but have issues particularly with young players in euro pro leagues as there is no TOI adjustment and these players are generally getting far fewer minutes then their junior level counterparts.

Furthermore, most players transitioning form SHL to NHL are older, so while it does have some issues, I think that's cleaned up pretty well with the favorable projection of 33%.

Also keep in mind this basing Defencemen progress off scoring points, which is less then perfect, but still has a decent success rate.

There are obvious imperfections, but there's a LOT of ground those imperfections would have to cover up to make it close, as Roslovic projected 2X as likely to succeed through pretty much every set of criteria I used. I don't think you can explain away a gap like that.

As for the D+1 season, i'll have to take a look. It's not done yet but 14 points 41 SHL games is not likely to have closed the gap considerably but will definitly be worth revisiting at the end of this year.

Thanks Grind.
I stand corrected then.

 

[Jet]

Best way to get guys who are not on the ice for goals against?
Get guys who are not on the ice for shots against...

Sorry, that's horse hockey.

I cannot control what the other 4 people on the ice with me do. There are shifts when I don't really do much out there because I have positive possession players on the ice with me. I play my role, retrieve the puck and get it into those players hands as soon as possible.

We also have several players on my team who struggle with possession. They aren't great at taking passes and are more likely to give the puck back to the opponent than make an offensive play. That can hardly be put on me. In those cases I will make safer plays: I will skate the puck out of my zone and then try to pass to them, or failing that, I will get accross the red line and pump it in. Occasionally I will attack the net but that's not my strength nor my desire.

When you are on the ice with poor possession players, guess what, you have the puck less. Hence you are probably in your zone more, and yes, maybe even blocking shots (I know it's a sin, but yes )

Hockey is a team sport. For the very reason people bemoan +/- we should be shouldn't have blanket thinking about possession and shot blocking.

And yes, again, call me a neanderthal, but I take pride in my +/- when I play. Friday night we lost 7 - 2 and I was a -1 with 2 D pairs. I am proud of that.

 

[Jet]

You describe a pretty good defenseman.

I think the bolded misstates the issue though. It seems to me that there is a misunderstanding between those who prefer that type of D man and those who prefer the modern offensive type. The offensive D man is presumed to be weak defensively allowing more goals against. He compensates for this weakness by scoring more goals as well. While that may happen sometimes I don't think that is the point or the objective. I think the idea is that the offensive D man has more puck possession which equals fewer SA against, therefore fewer goals against not more. He also scores more goals increasing his goal differential further. To take it a step further he also gets more O zone time which also reduces goals against and increases goals for. If he has fewer blocks it may be because he gives up fewer SAs. If he is weak at some of the more traditional defensive play he is making up for it by all those other things.

Lastly I don't think it is necessarily correct to say that either style is superior. It is the end result that matters and that is determined by the net goal differential at the end of the game. If one tends to win by 2-1 and the other tends to win 4-3 they are equal.

I wouldn't totally disagree with this statement. However, I think the role of a defenseman is first and foremost preventing goals (not simply outscoring). Why? Because I think in order to have a balanced team and to be able to play the situational game properly you need to have players who excel at a certain skill. If you are down a goal late in a game you need a guy who can put the puck in the net. If you are protecting a lead late in a game, I want a guy out there who's only passion in hockey is making forwards frustrated with superior positioning, stick use, agility and strength. I want a guy who smiles when a forward tries to double toe drag him and he just stays in the lane and puts him on his ass.

I'm not going to AGAIN enter the Buff debate here but it was no surprise to me when we got scored on in the last 30 seconds of the NYI game after just getting the lead. Was it Byfuglien's fault, specifically? No, not at all. But besides Toby Enstrom, who do we have out there that we feel our zone is pretty much on lock down when they are out there? Trouba is probably the closest guy to it.

I'd rather have 6 good puck movers who get the puck to our forwards, jump in occasionally, but always remember goal prevention is their primary role.

 

[Holden Caulfield]

Sorry, that's horse hockey.

I cannot control what the other 4 people on the ice with me do. There are shifts when I don't really do much out there because I have positive possession players on the ice with me. I play my role, retrieve the puck and get it into those players hands as soon as possible.

We also have several players on my team who struggle with possession. They aren't great at taking passes and are more likely to give the puck back to the opponent than make an offensive play. That can hardly be put on me. In those cases I will make safer plays: I will skate the puck out of my zone and then try to pass to them, or failing that, I will get accross the red line and pump it in. Occasionally I will attack the net but that's not my strength nor my desire.

When you are on the ice with poor possession players, guess what, you have the puck less. Hence you are probably in your zone more, and yes, maybe even blocking shots (I know it's a sin, but yes )

Hockey is a team sport. For the very reason people bemoan +/- we should be shouldn't have blanket thinking about possession and shot blocking.

And yes, again, call me a neanderthal, but I take pride in my +/- when I play. Friday night we lost 7 - 2 and I was a -1 with 2 D pairs. I am proud of that.

Which is why no one in their right mind talks about sample size at the one shift level

Your confusing individual situations with large sample size talk. Of course blocking shots is a good thing in terms of individual play. But it becomes a bad thing when you* are constantly forced to block shots since you are always in the your own end. Of course it happens you are on the ice with bad players and it's not your fault sometimes to get trapped in d-zone. However when you take that over the long term and you've played with everyone and it's still YOU that's being trapped, then your the bad player pulling people down, not your teammates.


Defensive defensemen are still GREATLY valued by both NHL GM's and the advanced stats community. The reality is the role of defensive defenseman has changed somewhat from big bruising punisher to positionally sound guy that can get the puck out of the zone, but it's still exists. What you are talking about is a guy like Tanev or Vlasic, those guys are much loved. Make smart safe plays with the puck, in position, and move the puck up the ice quickly and safely. But it's important to remember that a metric that is good at small sample sizes for analyzing a play or game is not the same as a metric that is good at analyzing big sample sizes.

*=Please note I am not saying "you" as in literally you, but you as is any random player

 

[Jet]

Which is why no one in their right mind talks about sample size at the one shift level

Your confusing individual situations with large sample size talk. Of course blocking shots is a good thing in terms of individual play. But it becomes a bad thing when you are constantly forced to block shots since you are always in the your own end. Of course it happens you are on the ice with bad players and it's not your fault sometimes to get trapped in d-zone. However when you take that over the long term and you've played with everyone and it's still YOU that's being trapped, then your the bad player pulling people down, not your teammates.

Defensive defensemen are still GREATLY valued by both NHL GM's and the advanced stats community. The reality is the role of defensive defenseman has changed somewhat from big bruising punisher to positionally sound guy that can get the puck out of the zone, but it's still exists. What you are talking about is a guy like Tanev or Vlasic, those guys are much loved. Make smart safe plays with the puck, in position, and move the puck up the ice quickly and safely. But it's important to remember that a metric that is good at small sample sizes for analyzing a play or game is not the same as a metric that is good at analyzing big sample sizes.

I agree with that, Holden. I was more railing against the very vague statement that garret made.

 

[garret9]

I'm not sure I am following this correctly. Do you mean that in exchange for a 3rd we should get an 8th + a 25th? ........ Or alternatively an 8th + 94 7ths?

The 94 7th round picks was a bit of a joke, but ya the other two are typical for trade value given NHL history.

 

[garret9]

Sorry, that's horse hockey.

I cannot control what the other 4 people on the ice with me do. There are shifts when I don't really do much out there because I have positive possession players on the ice with me. I play my role, retrieve the puck and get it into those players hands as soon as possible.

We also have several players on my team who struggle with possession. They aren't great at taking passes and are more likely to give the puck back to the opponent than make an offensive play. That can hardly be put on me. In those cases I will make safer plays: I will skate the puck out of my zone and then try to pass to them, or failing that, I will get accross the red line and pump it in. Occasionally I will attack the net but that's not my strength nor my desire.

When you are on the ice with poor possession players, guess what, you have the puck less. Hence you are probably in your zone more, and yes, maybe even blocking shots (I know it's a sin, but yes )

Hockey is a team sport. For the very reason people bemoan +/- we should be shouldn't have blanket thinking about possession and shot blocking.

And yes, again, call me a neanderthal, but I take pride in my +/- when I play. Friday night we lost 7 - 2 and I was a -1 with 2 D pairs. I am proud of that.

1) No that is hockey. And I'm not sure you understand what I'm getting at.
The best way to predict guys who don't have goals against is guys without shots against.
That's why we care about shots, because their predictive nature to goals.

2) NHL hockey rules and relationships we observe exist in the environment that is present. You can't compare other non-NHL levels. Heck, in only as far down as CHL, Corsi is much, much weaker and goal differentials are much, much stronger due to the differences in environment.
It will be even more extreme for level like beer league.

3) Corsi isn't possession. Possession and Corsi are different, as are shots. Possession with the absence of shots and opportunity to score does not create a shot event. You can see this in the extremes, like in soccer, which has far more dead possession time. It has a relationship, but it is not the same as Corsi. What Corsi is is a goal differential predictor.

4) I bemoan +/- for more reasons than just it being about goals. I bemoan it because it sucks.
* It is a raw +/- with no context of rates or TOI, so +5/-3 is same as +305/-303
* It includes empty nets, pp -s, and pk +s
It's a ****** stat. If you want to use goals, use even strength goal% (goals for/[goals against + goals against]). It is sensitive to TOI differences, and doesn't have the stupid PP/PK/EmptyNet rules. It still requires about 3-4 years to stabalize for players, but it's better.

 

[garret9]

Dustin Byfuglien, good defensive defenseman:

 

[garret9]

"Blocking shots is like killing rats. It's a good thing to do, but if you are doing it all the time it's a sign of other bad things."

This is true of almost any defensive play. We remember the plays a defenseman makes, but we don't understand how often they make the plays are in part due to their own play.

 

[Aavco Cup]

You are right. We are going around in circles. You seem to be willing to trust the Jets list without question.

I don't believe in positional drafting. Neither do I believe that anybody's evaluation is able to pick the absolute best prospect out of a group of 3-4 players taken 1 after the other in the draft after the first few. Even there we often see 3 or 4 players bunched so closely together that no one can really know which one is the best. Even among the experts you get many different opinions, all well thought out and defended by fact. They are looking at 17 YO kids and trying to identify who will be the best players 5-6 years later. So at each pick take the best player in the context of those available and of the team you are trying to build. Position is a factor 2 ways. Cs and Ds are more valuable than wingers and when you are loaded with Fs you should take the D out of that group of players who are so closely ranked that you really can't identify the best one. If one is clearly better than the others take him.

If I had the 1st OA this year I would not take a D. Mathews is clearly the best player bet. Laine is probably 2nd. After that it begins to get muddy.

Just as I can't convince you to draft positionally (not that I'm trying to), you can't convince me that any team's scouts can accurately put the top 30 players in order from 1 - 30. The error bars are huge. So when well within the error bars consider position along with several other factors.

Do you think anyone could look back at a draft 7 or 8 years ago and come up with a definitive list of the top 30 players from that draft, in order? That would be a list that every knowledgeable person would agree with.

incorrect. I am unwilling to trust your judgement over any of the 30 teams judgement.

 

[garret9]

BPA is one thing.
Perception of BPA is another.

You can't fault someone for going for their perception of BPA as if they were not going for BPA and went for the guy for another reason (ex: VAN and Virtanen). You can fault them if you know there is an issue with their process in evaluating what is BPA.

So going for Roslovic because they thought he was BPA, if he turns out not to be BPA, is one type of error.
Going for Virtanen despite the scouts on the team not thinking he's BPA is a different type of error.

 

[Mathmew Purrrr Oh]

Dustin Byfuglien, good defensive defenseman:

jamming that gap 24/7

 

[Mortimer Snerd]

I wouldn't totally disagree with this statement. However, I think the role of a defenseman is first and foremost preventing goals (not simply outscoring). Why? Because I think in order to have a balanced team and to be able to play the situational game properly you need to have players who excel at a certain skill. If you are down a goal late in a game you need a guy who can put the puck in the net. If you are protecting a lead late in a game, I want a guy out there who's only passion in hockey is making forwards frustrated with superior positioning, stick use, agility and strength. I want a guy who smiles when a forward tries to double toe drag him and he just stays in the lane and puts him on his ass.

I'm not going to AGAIN enter the Buff debate here but it was no surprise to me when we got scored on in the last 30 seconds of the NYI game after just getting the lead. Was it Byfuglien's fault, specifically? No, not at all. But besides Toby Enstrom, who do we have out there that we feel our zone is pretty much on lock down when they are out there? Trouba is probably the closest guy to it.

I'd rather have 6 good puck movers who get the puck to our forwards, jump in occasionally, but always remember goal prevention is their primary role.

I get your point. I don't care how they keep the puck out of their own net though. If they do it by their play in their D zone that's fine. If they do it by keeping the puck in the O zone that is equally fine.

I disagree with the bold though. If I was going to say something like that I would say that I would like 2 of those in the lineup. But I really wouldn't say that at all. I would like 6 good possession D men who do a good job outscoring their opponents one way or another.

 

[garret9]

I don't care how the team wins, as long as it is sustainable; I just care that they do win...

Or at least I care more about substance than style.

 

[Mortimer Snerd]

BPA is one thing.
Perception of BPA is another.

You can't fault someone for going for their perception of BPA as if they were not going for BPA and went for the guy for another reason (ex: VAN and Virtanen). You can fault them if you know there is an issue with their process in evaluating what is BPA.

So going for Roslovic because they thought he was BPA, if he turns out not to be BPA, is one type of error.
Going for Virtanen despite the scouts on the team not thinking he's BPA is a different type of error.

I agree with that. However I maintain that there is an issue with EVERYONE'S evaluation of who is the BPA at any point in the draft. It is not that they are incompetent. It is that no one can accurately predict who the best NHL players will be among a group of 17-18 year old kids. They can evaluate better than random choice but they cannot be absolutely accurate. Knowing this and knowing that a given subset of players are evaluated very close to one another you should know not to put excessive weight on those particular evaluations. That does not mean putting no weight on those evaluations. It simply means being aware that there is a high probability that the player you rate as 4th will turn out to be 5th and vice versa. Therefore while you generally subscribe to BPA it is permissible to also consider other factors. With the understanding that you also don't overweigh those other factors.

 

[csk]

I agree with that. However I maintain that there is an issue with EVERYONE'S evaluation of who is the BPA at any point in the draft. It is not that they are incompetent. It is that no one can accurately predict who the best NHL players will be among a group of 17-18 year old kids. They can evaluate better than random choice but they cannot be absolutely accurate. Knowing this and knowing that a given subset of players are evaluated very close to one another you should know not to put excessive weight on those particular evaluations. That does not mean putting no weight on those evaluations. It simply means being aware that there is a high probability that the player you rate as 4th will turn out to be 5th and vice versa. Therefore while you generally subscribe to BPA it is permissible to also consider other factors. With the understanding that also don't overweigh those factors.

So basically: tiers, then need within those tiers